The interest of researchers and organizations in learning at the workplace has been growing over the past years. The aim of this symposium is to widen our understanding of the learning that occurs at the workplace and to have more insight in the factors that contribute (or not) to learning at work. ‘Learning at work’ can be operationalised in different ways and different related factors can be taken into account. The four papers in this symposium all focus on individual and contextual factors that can explain differences in learning at the workplace. Each contribution has a different operationalisation of  ‘learning at work’ (self-directed learning in the first paper by Raemdonck, reflection in the workplace in the second paper by Kipfmueller and colleagues, work-related learning in the third paper by Gijbels and colleagues and deliberate practice in the fourth paper by Van de Wiel and colleagues). Also the nature of the workplace is different in the four papers (low qualified employees in the first paper, client advisors in the second study, ICT-workers in the third study and medical professionals in the fourth study). After the presentation of the papers the similarities, contradictions, gaps as well as opportunities of the results will be discussed by the discussant Lehtinen and with the audience.

This study examines the extent to which individual and company-level variables predict self-directedness in the learning of low-qualified employees. Multilevel analysis with a two-level approach was applied to examine which variables are significant predictors of self-directed learning. It was assumed that employees working in the same company are more alike in their self-directed learning than employees working in two different companies. Data was obtained from a unique sample of 408 low-qualified employees from 35 companies. The companies were selected from the energy sector, the chemical industry and the food industry.

The results show that the random effect of the intercept at company-level was not significant. The between company variance was attributed to the difference between the chemical industry and the other two sectors. The majority of the variance in self-directed learning was however explained by individual-level variables. Most striking was the key role of a proactive personality and past learning initiative. No significant cross-level interactions were found.

Introduction

Nowadays, self-directedness in learning and career processes is a key competence given the rapid rate of change, new technological opportunities, greater international competition, and increased flexibility in careers (Fay & Frese, 2001). However, in reality employees differ strongly to the extent in which they perform this key competence. Prior qualification level is often stated as an important predictor of level of self-directedness although this hypothesis is hardly based on empirical evidence. Differences among employees might be explained by other factors than educational qualification background (see study of Raemdonck, Thijssen & Valcke, 2005). If low-qualified employees do not significantly differ from high-qualified employees, there must be other reasons why some low-qualified workers succeed in self-directing their learning and career processes and other low-qualified workers do not. What factors lead some low-qualified workers to be more self-directed in their learning and career processes than other low-qualified employees? This is the central research question in this paper presentation. On the base of a review of the literature, a theoretical model was developed and tested. The model distinguishes between individual level factors (demografics, personality variables, human capital variables, job characteristics) and organizational level factors (economical sector, size, staff policy).

Method

408 low-qualified employees from 35 companies located in Flanders were individually interviewed by means of a questionnaire. The companies were selected from the energy sector, the chemical industry and the food and nourishment industry. The selection of the organizations was based on the BELFIRST-database. All organizations with ≥ 20 employees and NACE-BEL code 15, 24, 40 and 41 were contacted. In each organization, at least 10 low-qualified employees were selected at random. ‘Low-qualified’ was operationalised as a person with no diploma of initial secondary education. Prior to the interview, each participant signed an informed consent form.

Analyses

Multilevel regression modeling was applied to test our hypothesis with regard to the prediction of the two dependent variables: self-directedness in learning and career respectively. The hierarchical structure of the data suggests multilevel regression analysis approach. Employees were nest within companies (representing cultures and contexts), and companies are nested within sectors (chemical industry, energy sector and food and nourishment industry). Theoretically, this data structure yields three levels. However, because at the sector level only three units were observed, it was decided to apply a two-level approach with employees at level-1 nested within companies at level-2. In these models sector was included as a between company variable (at level-2).

Results

Predictors of self-directedness in learning processes (SDL)

The unconditional model for SDL yielded an intra-class correlation estimate of .05. Extension of this model by incorporating dummy variables coding the different sectors showed that a significant amount of the between company variance was explained by the difference between the chemical industry and the other sectors. Conditional to this variable the intra-class correlation dropped to .02. In subsequent models sector differences were therefore included by adding a dummy variable indicating the difference between the chemical industry and other sectors (CHEMDUM). Differences between the other sectors did not show significant effects.

Next, predictor variables and interaction variables were added to the model. For interaction variables no significant fixed effects were found.

Not surprisingly, because of the relatively low conditional intra-class correlation, the random effect of the intercept at Level-2 was not significant (χ2 = 1.94, df = 1, p > .05). None of the predictor variables showed a significant random effect (random slope) as well.

Significant positive effects on employees` level of self-directedness in learning processes are shown for the variables proactive personality, pursuit of knowledge work, past learning initiative, task variety and growth potential of the job. Concerning company characteristics significant positive effects were found for the variables participatory staff policy and CHEMDUM (chemical industry versus the other two sectors)

Predictors of self-directedness in career processes (SDC)

The unconditional model for SDC yielded an intra-class correlation estimate of .14. Similar to SDL, only the dummy variable indicating the difference between chemical industry and the other sectors showed a significant effect (explained a significant amount of the between company variance). Conditional to this variable the intra-class correlation dropped to .06. 

Next, predictor variables and interaction variables were added to the model. No significant fixed effects of interaction variables were found. For SDC the random effect of the intercept at Level-2 was significant (χ2 = 6.16, df = 1, p < .02). For the predictor variables no random effects (random slopes) were found.

Significant positive effects on employees` level of self-directedness in career processes are shown for the variables proactive personality, pursuit of knowledge work, mobility aspiration (striving for higher job position versus not striving for higher job position), and growth potential of the job. A significant negative effect was found for the variable age. Concerning company characteristics the only significant, positive effect was found for the variable CHEMDUM.

Conclusion

Differences in self-directedness in learning and career processes between low-qualified employees can be mainly explained by variables at the individual employee level. However, there are clear indications that the culture and context in which low-qualified employees are working do play a significant role. Especially in the prediction of self-directedness in career processes, the economical sector the company belongs too, does play an important role.

Literature

Fay, D., & Frese, M. (2001). The concept of personal initiative: An overview of validity studies. Human Performance, 14, 97-124.

Raemdonck, I., Thijssen, J., & Valcke, M. (2005). Self-directedness in learning and career processes. Lifelong Learning in Europe, 10(2), 76-81.

The study investigated different attitudes towards errors as predictors of reflection at work. Organisational climate for psychological safety was assumed to have a mediating influence. 84 client advisors in retail banking departments were questioned. Their workplaces were affected by different workplace changes by the time the data was collected. Error competence, learning from errors, and error strain showed to be strong predictors for reflection. A mediating role of psychological safety on the identified influences could be confirmed for the relationship to colleagues and to supervisors.

Errors in the workplace stand as work life instances requiring directed and effective action as they are often being labeled as critical and challenging. Common definitions label errors as deviations from individual expectations in terms of anticipated or desired results. This view already contains a strong consonance with reflection as it is commonly conceptualised as a post-event examination of individual’s own actions. It can include reappraisal and interpretation of experienced phenomena, an analysis of their causes and effects and drawing conclusions for actions in the future (Boud et al., 1989). Although errors can interfere with work processes, they can serve as important opportunities for learning (Gartmeier et al., 2008).

The study’s primary aim was to analyse the interrelation between attitudes towards errors and reflection (aim A). Furthermore it was examined to what extend the working climate within the team and with regard to supervisors (psychological safety) has a mediating effect on the above mentioned coherency (aim B).

Methodology

Sample

A sample of N=84 client advisors (45% male, 50% female, 5% not reported) voluntarily filled in questionnaires directly at their workplaces.

Instruments

Throughout the whole questionnaire, a six-type Likert scale was used ranging from 1 = totally agree up to 6 = totally disagree.

Error orientation. Attitudes towards errors at work were assessed by five scales of the error orientation questionnaire (Rybowiak et al., 1999): Error competence (four items, α = .80, M = 2.06, SD = 0.58), learning from errors (four items, α = .86, M = 2.36, SD = 0.78), error strain (five items, α =.63, M = 4.17, SD = 0.74), error anticipation (five items, α = .66, M = 3.02, SD = 0.72), and error risk taking (four items, α = .77, M = 2.77, SD = 0.83).

Reflection. An instrument of Kauffeld (2007) was applied to assess reflection at work (four items, α = .84, M = 2.33, SD = 0.63).

Psychological safety. In order to measure the perceived working climate (psychological safety), items of established instruments (Edmondson 1999; Van Dyck et al., 2005) were adapted (psychological safety colleagues: four items, α = .78, M = 2.44, SD = 0.70; supervisors: four items, α = .73, M = 2.41, SD = 0.64).

Findings

Correlation analyses showed significant interrelations between reflection and three facets of error orientation. In detail, reflection is significantly positive correlated with error competence as well as learning from errors, and negatively correlated with error strain (table 1).

Table 1. Correlation analyses

Notes. * = p < .05; ** = p < .01.

Further, it was analysed whether the relation between error orientation and reflection was mediated by psychological safety. Both psychological safety – colleagues and psychological safety – supervisors were found to be significant mediators in a number of aspects (table 2).

Table 2. Regression analyses for mediator effects

Notes. *p < .05, **p < .01; CPT = error competence, LRN = learning from errors, RSK = error risk taking, STR = error strain, ACP = error anticipation, PS-C = sychological safety colleagues, PS-S = psychological safety supervisors, REF = Reflection.

Discussion

When trying to understand learning through reflection at work individual factors like attitudes towards errors as well as contextual factors like the working climate (psychological safety) have to be taken into account. According to the results, the individual’s conviction to successfully cope with errors (error competence) as well as the belief in the learning potential of errors (learning from errors) obviously has positive effects on the willingness to critically think about one’s own actions at work. The perception of making an error as a strain obviously leads to a decreased willingness to reflect.

Regarding the mediating role of psychological safety it can be assumed that a working climate within the team and with regard to the supervisors which is characterised through an open communication about errors and a cooperative handling of occurred errors encourages reflection. A working climate which is based on mutual trust, assistant co-operation, and constructive communication supports the effects of attitudes towards errors on reflection. Through that reflective activities allow error-related learning at work. 

Based on the Demand-Control-Support (DCS) model (Johnson & Hall, 1988; Karasek, 1979; Karasek & Theorell, 1990) and the research by Raemdonck (2006) the present paper aims to investigate the influence of job-characteristics such as job demands, job control, social support at work on the one hand and self-directed learning orientation on the other hand on the work-related learning behaviour of the worker.  The study took place in the ICT-department of a large company in Flanders. By means of an online questionnaire all employees of the ICT-department were asked to complete a questionnaire which consisted apart from general information on the participants (age, sex, prior education etc.) on 5 scales (job demands, job control, social support, self-directed learning orientation and work-related learning behaviour) adapted from validated instruments. A total of 73 participants participated in the research  (responsrate of 52% ,    73% man, 27% women, age varieng from 20-51 years old). All scales had Cronbach’s alpha values above .79.  The results of the linear regression analyses indicated that the self-directed learning orientation scale predicted the work-related learning behaviour significantly. Also relevant prior education and high job demands were linked to work-related learning behaviour. Several explanations are discussed in the paper.

Aims

Based on the Demand-Control-Support (DCS) model (Johnson & Hall, 1988; Karasek, 1979; Karasek & Theorell, 1990) and the research by Raemdonck (2006) the present paper aims to investigate the influence of job-characteristics such as job demands, job control, social support at work on the one hand and self-directed learning orientation on the other hand on the work-related learning behaviour of the worker. 

Methodology

The present study took place at  Kluwer Technology Solutions (KTS). KTS is part of Kluwer Belgium,  a leader in information services for professionals in different areas A total of 73 participants  (response rate of 52% ). 73%  of the participants was male,  27% was female. The age of the participants varied between 20-51 years old.

The research instrument was administered online and consisted of different scales based on existing and prior validated questionnaires. The scales for ‘self-directed learning orientation (SDLO)’ and ‘work-related learning’ ware based on the research by Raemdonck (2006).  The scales for ‘job demands’ and for ‘job control’ were based on the research by Hackman & Oldham (1980), Krasek (1985),  Taris and Kompier (2004) and  De Jonge and colleagues (2000). Finally, the scale for ‘social support’ was based on the research by Van Veldhoven and colleagues (2002). All scales had Cronbach alpha scores higher than 0 .79. Also some demographic Information about the respondents, such as their age, sex, relevant prior education, prior experiences and aspirations, was gathered in the questionnaire as a source of control.

Findings

The regression-analysis, including all variables in the model, explains 38% of the variance within work-related learning behaviour.  However, only SDLO (β-= 0 .473, p<0.001)  and relevant prior education (β-= 0 .213, p<0.05)  were responsible for the significant effect (see Table 1).    

Table 1

Constant                                           -.353                             .651                           

Job demands                                      .157                            .134                            .149     

Job control                                         .103                            .093                            .126     

Social support                                              -.013                             .091                           -.015

SDLO                                                .569                            .132                             .473** *

Relevant prior knowledge                 .279                            .140                            .213*

Experience                                         .001                            .001                            .162

Sex                                                     .024                            .122                            .020

            Adj. R² = .383                                                * p < .05         ** p < .01       *** p < .001

In a next model (see Table 2), only ‘relevant prior education’ is taken into account as a control variable. In this model, we find an adjusted R² of .432.  SDLO remains significant (β-= 0 .474, p<0.001) and also relevant prior Education remains significant (β-= 0 .224,  p>0.05). Remarkable is the effect of job demands, which is now close to significant (β-= 0 .163, p=0.087).

Table 2

Constant                                           -.280                             .531                           

Job demands                                      .172                            .099                            .163     

Job control                                         .075                            .081                            .092     

Social support                                              -.024                             .077                           -.029

SDLO                                                 .571                            .119                             .474** *

            Adj. R² = .432                                                * p < .05         ** p < .01       *** p < .001

            remark: p (job demands) < .10

Based on these results, we build a new model (see Table 3) with SDLO, job demands and relevant prior education as the sole variables. This model explains 44% of the variance, with all three variables having a significant positive effect on the depended variable work-related learning. SDLO seems to have the largest effect (β-= 0 .491, p<0.001).

Table 3

Constant                                           -.294                             .480                           

Job demands                                      .193                            .096                            .184*   

SDLO                                                 .591                            .114                             .491** *

            Adj. R² = .432                                                * p < .05         ** p < .01       *** p < .001

Discussion

The results of the linear regression analyses indicated that the self-directed learning orientation scale predicted the work-related learning behaviour significantly. Also relevant prior education and high job demands were linked to work-related learning behaviour.  It is remarkable that variables related to the person had most effect on the work-related learning behaviour of the respondents. The impact of work-related variables seems rather modest, only high job demands seemed to have a positive effect. Investing in a self-regulated learning orientation of workers seems to be most important to enhance work-related learning.

This study investigated the workplace learning of residents in internal medicine. Residents are trained as medical professionals on the job while working in clinical practice. However, it is argued that to benefit from these work experiences, deliberate practice is required, i.e., a focused effort on the development of performance aspects that need improvement. The present study questioned in what way residents of internal medicine engage in deliberate practice at the workplace, what factors contribute to that, and how this affects expertise development.

We hypothesized that an active learning attitude focused on the improvement of performance leads to engagement in meaningful learning activities (i.e. deliberate practice) which contribute to the development of medical expertise. We further hypothesized that work experiences and work environment would influence these processes. To test these hypotheses, a questionnaire measuring residents’ behavior and attitudes in their work and training, including items on their experience and work environment, as well as a case test measuring expertise, was completed by 37 residents. The results showed that residents had overall a high focus on learning. Moreover, mastery goals predicted the engagement in learning activities, whereas performance avoidance goals had a negative influence. However, engagement in learning activities was not related to performance on the case test, whereas experiences in affiliated hospitals in and before residency were. Work environment factors seemed to play an important role in shaping residents’ experiences and producing expertise. We conclude that work in affiliated hospitals is useful for building clinical knowledge, but may be harmful for active learning attitudes and behaviors. The results provide suggestions for organizing the workplace of residents in such a way that their learning is fostered. Moreover, this study shows the potential of using deliberate practice theory to capture expertise development in the workplace.

In becoming a medical professional, residents need to combine clinical practice with learning on the job. In their work the diagnosis and treatment of patients is central. They may learn from the problems encountered, the feedback generated by diagnostic tests, the course of illness, and consultations with supervisors and other colleagues, as well as from self-study, seminars, courses and lectures. However, it is argued that to benefit from these experiences, deliberate practice is required, i.e., a focused effort on the development of performance aspects that need improvement (Ericsson, 2004). This process requires high level self-regulated learning skills (Zimmerman, 2006), as well as the motivation to improve competence, i.e., a mastery goal orientation (Pintrich, 2000).

In the present study we questioned in what way residents of internal medicine engage in deliberate practice at the workplace, what factors contribute to that, and how this affects expertise development. More specifically, we hypothesized that mastery goals will lead to relevant learning activities at the workplace. These learning activities, combined  with experience in diagnosing and treating patients will promote the development of expertise. We further hypothesized that a performance-avoidance goal orientation, i.e., the tendency to avoid failure and unfavorable judgments, hinders expertise development, as relevant deliberate practice activities, such as consulting knowledgeable colleagues and seeking accurate feedback, are avoided. Work environment variables, such as psychological safety, and work setting (academic versus affiliated hospitals) may stimulate certain goal orientations and provide specific opportunities for practice that shape the way residents learn from their experiences. We supposed that an environment which is not regarded as psychological safe will foster performance avoidance goals, leading to less engagement in deliberate practice.

Method

Based on an interview study with internists (Van de Wiel et al., 2007), we developed a questionnaire to measure the behavior and attitudes of residents in their work and training. We focused on their goals, problem solving strategies, advice and feedback seeking, reflection, planning of learning and updating activities, as well as on their work and educational environment. A case test, composed of 27 items of the  Medical Knowledge Self Assessment Program (MKSAP) 2004-2005, was used to asses medical expertise. Participants were 37 residents in internal medicine working in academic and affiliated hospitals.

 Results

The results showed an overall focus on learning by residents as indicated by high mastery goal orientations (M = 5.57; SD = .63; Cronbach alpha = .72) and low performance avoidance goal orientations (M = 2.78; SD = 1.01; Cronbach alpha = .76), and moderate to high levels of deliberate practice for most identified work-related learning activities (Table 1).

Table 1. Descriptive results on the learning activities and attitudes categories of the deliberate practice questionnaire.

* When a scale’s Cronbach alpha < .48 we used the most outspoken deliberate practice item.

** 1-7 Likert scale response format

Mastery goals predicted the engagement in learning activities, whereas performance avoidance goals had a negative influence (Table 2). However, engagement in learning activities was not related to performance on the case test (r = -.044, p = .80), whereas experiences in affiliated hospitals in and before residency were (respectively, r = .38, p = .019, and r = .40, p = .027; Table 3).

Table 2. Regression of mean deliberate practice on goal orientations.

* p ≤ .1; ** p ≤ .05; *** p ≤ .001

Table 3. Regression of case test on mean deliberate practice and two experience variables.

* p ≤ .1

Work environment factors seemed to play an important role in shaping residents’ experiences and producing expertise. Generally, residents described their workplace as psychologically safe (M = 5.56; SD = .69; Cronbach alpha = .65). These ratings were positively correlated with engagement in deliberate practice (r = .30, p = .075), but negatively with performance avoidance goals (r = -.50, p = .002). Experience on the wards before official training, however, was positively correlated with performance avoidance goals (r = .37, p = .040). Experience in affiliated hospitals and before residency, furthermore, were negatively correlated with seeking feedback (respectively, r = -.46, p = .004, and r = -.36, p = .049). Reflection, in addition, was negatively correlated with working hours per week (r = -.33, p = .050), and also with case test performance (r = -.36, p = .029).

Discussion

The data provided support for most of our hypotheses. Surprisingly, however, there were no indications that relevant learning activities for practicing medicine based on deliberate practice theory and theories of self-regulated learning contributed to performance on the case test. Reflection was even negatively correlated with the expertise measure.

This last finding may indicate that those with the least experience recognize that they need deliberate practice so they invest in activities that contribute to their professional development. A complementary explanation is that work in affiliated hospitals and before residency is useful for building clinical knowledge, but at same time may be harmful for an open learning attitude. The study provides suggestions for organizing the workplace of residents in such a way that their learning is fostered. Moreover, it shows the potential of using deliberate practice theory to capture expertise development in the workplace.

At this moment a follow-up study is conducted. Results will be available at the conference.